Eye tracking may reduce cases of breast cancer misdiagnosis

It’s estimated by the American Cancer Society that over 229,000 new cases of breast cancer will be diagnosed this year. Even with advances in mammography technology, a portion of these cases will be misdiagnosed, causing patients to endure unnecessary treatments and psychological damage.

Modern imaging technology is able to produce higher resolution images that reveal smaller lesions on which pathologists now must make judgment calls. Determining whether or not a lesion is benign or early stage cancer is a particularly challenging decision to make. Cindy M. Grimm, PhD, associate professor of computer science and engineering at Washington University, and Reynold Bailey, PhD, assistant professor of computer science at the Rochester Institute of Technology, conducted an eye tracking study that suggests using the scanpaths of expert radiologists may be an effective training tool for novices.

An expert diagnostic radiologist was hired to analyze 65 mammogram images from the Mammographic Image Analysis Society database. Using an eye tracking system, the expert’s scanpath was recorded to create pattern to be tested with novices. A technique developed by Grimm and Bailey, called subtle gaze direction, was implemented to guide the novices’ focus by changing the brightness along the expert’s scanpath pattern. The stimulus appeared in the peripheral field of view, directing the viewer’s scanning along the specified path but was cut off before it could be focused on so as to alter the viewing experience as little as possible.

The results revealed that the novices who were guided along the expert scanpath with subtle gaze direction were significantly more accurate at identifying cancerous lesions than both the control group that had no guidance and a third group that was guided on a random path. In addition, it was found that the novices who were trained with the expert scanpath remembered the technique, and its effectiveness was carried on beyond the study. This has promising implications that suggest a training of this sort could be developed and implemented to enhance the skills novices in reading mammograms and many other forms of medical imaging.

This expert versus novice application is not the first we’ve seen here at Eye Tracking Update. In fact, it seems to be a trend that we will be hearing about more of in the future. Check out a few other studies that have been done using similar eye tracking tactics:
Eye tracking study may reduce teen crash rates
Eye Tracking Shows There is No Horsing Around in Show Jumping
Eye Tracking: How Experience is Key in Becoming an Expert
For more information on this Washington University study, visit http://news.wustl.edu/news/Pages/23315.aspx

No related articles.